Methods and Applications of Computational Design in Multiple States

• Main Author: Dowling, Quinton M
• Format: Dissertation
• Language: English
• Published: ProQuest Dissertations & Theses 01-01-2023
• Subjects: Biochemistry; Bioengineering; Materials science; Nanoscience
• ISBN: 9798379907624

The balance between flexibility and stability is a key property of proteins, contributing to the vast area of structures and functions observed in nature. Flexibility arises from the complex energy landscape that determines the available conformational states of a protein. Over the past two decades, our ability to design proteins has advanced significantly, however, our understanding of and ability to design multistate proteins remains an area of significant challenge. The multistate design challenge hinders our ability to engineer functional proteins like enzymes, and limits the size and complexity of designed protein materials. The focus of my doctoral research has been developing multistate design approaches and applying those methods to practical design problems; a peudosymmetric trimer used as a building block for assembling large, closed, cage-like nanostructures; and re-engineering a natural enzyme to remove allosteric dependence on double-stranded DNA for activity. In both cases I started by applying simple computational methods to identify mutations that are likely to shift the equilibrium from an undesired state to a desired state. I then incorporated bioinformatics data to improve the design pipeline. This approach was applied to convert a naturally occurring, symmetric, homotrimeric protein, 1WA3, into a pseudosymmetric heterotrimer. I used the resulting pseudosymmetric trimer as a building block for designing cage-like protein assemblies. Because of the choice of trimer as starting material I was able to efficiently build cage-like structures containing 240, 540, or 960 protein chains, significantly larger than any previous computationally designed, bounded, protein nanostructure. I also applied this multistate design approach to a naturally occurring enzyme, cyclic GMP-AMP synthase (cGAS). Under normal circumstances cGAS adopts an enzymatically active conformation only when bound to double-stranded DNA (dsDNA). By applying multistate design I was able to engineer constitutively active variants of cGAS, which adopt the active conformation independent of dsDNA. We then showed the utility of CA-cGAS in an in vivo model. The methods developed here are generally applicable to multistate design problems in naturally occurring building blocks and also highlight the practical utility of the proteins engineered using these approaches.